CaPTk

CaPTk analyzes radiographic cancer images to quantify morphology and function from biomedical images and to link imaging phenotypes with genetic and molecular characterizations for predictive radiomic and radiogenomic modeling.

Key Features:

• Cancer focus: Supports analysis of brain (including glioblastoma multiforme), breast, and lung cancers.
• Quantitative morpho-functional analysis: Quantifies morphology and functional information from biomedical imaging data.
• Multiparametric MRI feature extraction: Extracts comprehensive imaging features from presurgical multiparametric MRI (MP-MRI) scans.
• Radiomic model construction: Constructs radiomic models and develops predictive imaging signatures across single- and multi-institutional datasets.
• Radiogenomic integration: Links imaging-derived phenotypes with genetic and molecular characterizations.
• Advanced computational techniques: Employs advanced computational techniques to derive imaging signatures predictive of clinical outcomes.
• Model evaluation metrics: Supports evaluation using area under the receiver operating characteristic curve (AUC).

Scientific Applications:

• GBM prognostic modeling: Used to develop predictive signatures for progression-free survival (PFS) and recurrence pattern (RP) from presurgical MP-MRI in patients who underwent gross total resection followed by standard adjuvant chemoradiation therapy, with reported AUCs of 0.88 (single-institution) and 0.82–0.83 (multi-institutional) for PFS and 0.88 (single-institution) and 0.56–0.71 (multi-institutional) for RP.
• Radiogenomic biomarker development: Integrates imaging phenotypes with genetic and molecular data to support diagnostic and predictive biomarker discovery.
• Multi-institutional validation: Enables construction and validation of radiomic signatures across institutions to assess generalizability.

Methodology:

Extraction of comprehensive imaging features from presurgical MP-MRI scans; construction of radiomic predictive signatures/models for progression-free survival and recurrence pattern; multi-institutional model development and evaluation using area under the receiver operating characteristic curve (AUC); linkage of imaging phenotypes with genetic and molecular characterizations.